Coal preparation plant personnel management platform and method based on UWB positioning

Abstract

The invention relates to the field of personnel management, and particularly discloses a UWB positioning-based coal preparation plant personnel management platform and method, and the method comprises the steps: firstly obtaining coal preparation plant personnel positioning data collected by a UWB positioner, then carrying out the feature extraction and correlation analysis of the data through a deep learning technology, and finally carrying out the classification of the data through a classifier, therefore, potential safety accidents can be effectively prevented, the management cost is reduced, and the safety, the production efficiency and the management level of the staff of the coal preparation plant are further improved.

Description

Technical Field

[0001] The present application relates to the field of personnel management, and more specifically, to a personnel management platform and method for coal preparation plants based on UWB positioning. Background Art

[0002] A coal preparation plant is a factory that removes impurities from raw coal and extracts high-quality coal through a series of physical and chemical methods. Its main processes include crushing, screening, gravity separation, flotation, etc. Through these processes, coal is separated from impurities such as stones and sediment to achieve the purpose of increasing the calorific value and quality of coal. Potential risks such as high temperature, toxic gases, and equipment failures usually exist in the dangerous areas of coal preparation plants, posing a threat to the lives and safety of employees. Therefore, it is necessary to manage the positioning of employees in coal preparation plants to monitor the positions of employees, ensure that each employee is working in the designated area, and prevent improper personnel flow or entry into dangerous areas.

[0003] General GPS or Beidou personnel positioning is applicable to scenarios with open and unobstructed areas and low positioning accuracy requirements. It has poor positioning accuracy for application scenarios such as coal preparation plants and equipment rooms where equipment is dense, satellite signals are blocked, and high positioning accuracy is required.

[0004] Therefore, a personnel management platform and method for coal preparation plants based on UWB positioning are desired. Summary of the Invention

[0005] To solve the above technical problems, the present application is proposed. Embodiments of the present application provide a personnel management platform and method for coal preparation plants based on UWB positioning. First, it obtains the positioning data of employees in coal preparation plants collected by UWB locators, then uses deep learning technology to perform feature extraction and correlation analysis on it, and finally passes it through a classifier to determine whether to issue an early warning alarm for abnormal personnel positioning, so as to effectively prevent potential safety accidents, reduce management costs, and thus improve the safety, production efficiency, and management level of employees in coal preparation plants.

[0006] According to one aspect of the present application, a personnel management platform for coal preparation plants based on UWB positioning is provided, which includes:

[0007] A module for obtaining data of employees in coal preparation plants, which is used to obtain the positioning data of employees in coal preparation plants collected by UWB locators;

[0008] A module for extracting data of employees in coal preparation plants, which is used to extract the feature vectors for understanding the positioning text of employees in coal preparation plants and the correlation feature vectors of the positioning data of employees in coal preparation plants from the positioning data of employees in coal preparation plants collected by the UWB locators;

[0009] A personnel positioning anomaly judgment module, which is used to judge whether to issue a personnel positioning anomaly warning alarm based on the coal preparation plant employee positioning text understanding feature vector and the coal preparation plant employee positioning data association feature vector.

[0010] According to another aspect of the present application, there is provided a personnel management system for a coal preparation plant based on UWB positioning, which includes:

[0011] Obtain the coal preparation plant employee positioning data collected by the UWB locator;

[0012] Extract the coal preparation plant employee positioning text understanding feature vector and the coal preparation plant employee positioning data association feature vector from the coal preparation plant employee positioning data collected by the UWB locator;

[0013] Based on the coal preparation plant employee positioning text understanding feature vector and the coal preparation plant employee positioning data association feature vector, judge whether to issue a personnel positioning anomaly warning alarm.

[0014] Compared with the prior art, the personnel management platform and method for a coal preparation plant based on UWB positioning provided by the present application first obtains the coal preparation plant employee positioning data collected by the UWB locator, then uses deep learning technology to perform feature extraction and correlation analysis on it, and finally uses a classifier to judge whether to issue a personnel positioning anomaly warning alarm, so as to effectively prevent potential safety accidents, reduce management costs, and further improve the safety, production efficiency and management level of the coal preparation plant employees. Description of the Drawings

[0015] By describing the embodiments of the present application in more detail in conjunction with the drawings, the above and other objects, features and advantages of the present application will become more obvious. The drawings are used to provide a further understanding of the embodiments of the present application, and constitute a part of the specification. Together with the embodiments of the present application, they are used to explain the present application and do not constitute a limitation to the present application. In the drawings, the same reference numerals generally represent the same components or steps.

[0016] Figure 1 It is a block diagram schematic diagram of a personnel management platform for a coal preparation plant based on UWB positioning according to an embodiment of the present application.

[0017] Figure 2 It is a block diagram schematic diagram of a coal preparation plant employee data extraction module in a personnel management platform for a coal preparation plant based on UWB positioning according to an embodiment of the present application.

[0018] Figure 3 It is a block diagram schematic diagram of a coal preparation plant employee positioning semantic encoding unit in a personnel management platform for a coal preparation plant based on UWB positioning according to an embodiment of the present application.

[0019] Figure 4It is a block diagram schematic of a personnel positioning anomaly judgment module in a coal preparation plant personnel management platform based on UWB positioning according to an embodiment of the present application.

[0020] Figure 5 It is a flowchart of coal preparation plant personnel management based on UWB positioning according to an embodiment of the present application. Detailed implementation manners

[0021] Various exemplary embodiments, features, and aspects of the present application will be described in detail below with reference to the accompanying drawings. The same reference numerals in the drawings denote elements having the same or similar functions. Although various aspects of the embodiments are shown in the drawings, the drawings do not have to be drawn to scale unless otherwise specified.

[0022] The special term "exemplary" here means "serving as an example, embodiment, or illustration". Any embodiment described as "exemplary" here does not have to be construed as superior to or better than other embodiments.

[0023] In addition, in order to better illustrate the present application, numerous specific details are given in the following detailed implementation manners. Those skilled in the art should understand that the present application can also be implemented without some specific details. In some instances, methods, means, elements, and circuits well known to those skilled in the art are not described in detail so as to highlight the gist of the present application.

[0024] Furthermore, the terms "first" and "second" are only used for descriptive purposes and cannot be construed as indicating or implying relative importance or implicitly specifying the quantity of the indicated technical features. Thus, features defined with "first" and "second" may explicitly or implicitly include one or more of such features. In the description of the present application, "a plurality of" means two or more unless otherwise specifically defined.

[0025] Figure 1 It is a block diagram schematic of a coal preparation plant personnel management platform based on UWB positioning according to an embodiment of the present application. As Figure 1 shown, the coal preparation plant personnel management platform 100 based on UWB positioning according to an embodiment of the present application includes: a coal preparation plant employee data acquisition module 110 for acquiring coal preparation plant employee positioning data collected by a UWB locator; a coal preparation plant employee data extraction module 120 for extracting a coal preparation plant employee positioning text understanding feature vector and a coal preparation plant employee positioning data association feature vector from the coal preparation plant employee positioning data collected by the UWB locator; and a personnel positioning anomaly judgment module 130 for judging whether to issue a personnel positioning anomaly warning alarm based on the coal preparation plant employee positioning text understanding feature vector and the coal preparation plant employee positioning data association feature vector.

[0026] In the above-mentioned personnel management platform 100 for coal preparation plants based on UWB positioning, the coal preparation plant employee data acquisition module 110 is used to acquire the positioning data of coal preparation plant employees collected by UWB locators. It should be understood that to acquire the positioning data of coal preparation plant employees collected by UWB locators, real-time monitoring is usually required through a UWB (Ultra-Wideband) positioning system. The UWB positioning technology installs a series of base stations or anchors inside the coal preparation plant. These base stations communicate with the employee positioning devices equipped with UWB tags through ultra-wideband wireless signals to obtain the location information of the employees. First, multiple base stations need to be set up in the coal preparation plant area. These base stations will conduct frequent data exchanges with the positioning tags on the employees. The base stations calculate the distance by measuring the signal transmission time and signal strength between the base stations and the tags, and finally achieve positioning. To accurately acquire the positioning data, these base stations are usually connected to the central control system. The central system will receive the data of all base stations in real time and calculate the accurate location of the employees through data fusion algorithms. These positioning data can be transmitted to the server or cloud platform through the wireless network for storage and analysis, helping to master the dynamics of the employees and ensure safe production. In addition, the UWB positioning system can also provide high-precision positioning information, effectively avoiding the errors of traditional positioning methods. Especially in complex environments such as industrial sites like coal preparation plants, it can ensure the accuracy and timeliness of the employee location data.

[0027] Specifically, considering that traditional GPS or Beidou positioning systems are usually applicable to open areas and environments with relatively low positioning accuracy requirements. However, in complex scenarios such as coal preparation plants where there are dense equipment and satellite signals are easily blocked, the positioning accuracy of these systems often fails to meet the high-demand application requirements. Therefore, in the technical solution of this application, by acquiring the positioning data of coal preparation plant employees collected by UWB locators and combining deep learning technology, it is possible to determine whether to issue an early warning alarm for abnormal personnel positioning, thereby effectively preventing potential safety accidents, reducing management costs, and further improving the safety, production efficiency, and management level of coal preparation plant employees.

[0028] In the above-mentioned personnel management platform 100 for coal preparation plants based on UWB positioning, the coal preparation plant employee data extraction module 120 is used to extract the coal preparation plant employee positioning text understanding feature vector and the coal preparation plant employee positioning data association feature vector from the positioning data of coal preparation plant employees collected by the UWB locators. It should be understood that by combining the coal preparation plant employee positioning text understanding feature vector and the coal preparation plant employee positioning data association feature vector, important information such as whether the employee is in a dangerous area and whether the employee works according to the predetermined process can be identified, which can be further used for applications such as safety monitoring, behavior prediction, and anomaly detection to ensure the safety and work efficiency of the employees.

[0029] Figure 2It is a block diagram schematic of the coal preparation plant employee data extraction module in the coal preparation plant personnel management platform based on UWB positioning according to an embodiment of the present application. As Figure 2 shown, in a specific embodiment of the present application, the coal preparation plant employee data extraction module 120 includes: a coal preparation plant employee positioning semantic encoding unit 121, configured to perform semantic encoding on the coal preparation plant employee positioning data collected by the UWB locator to obtain the coal preparation plant employee positioning text understanding feature vector; a coal preparation plant employee positioning feature encoding unit 122, configured to perform feature encoding on the coal preparation plant employee positioning data collected by the UWB locator to obtain the coal preparation plant employee positioning data association feature vector.

[0030] It should be understood that UWB positioning data includes information such as the location, movement trajectory, and residence time of employees. Although these data can provide basic spatial positions themselves, they often lack sufficient semantic depth and are difficult to directly reflect the behavior intentions, work tasks, or potential safety risks of employees. Through semantic encoding, these data can be transformed into more meaningful feature vectors to help the system understand the behavior patterns and working environments of employees. This semantic encoding can not only enhance the expression ability of positioning data but also enable the system to better combine environmental information to judge whether the behavior of employees complies with safety regulations or predetermined work processes when dealing with complex scenarios.

[0031] Furthermore, the text understanding feature vector obtained solely by relying on semantic encoding still cannot comprehensively capture the behavior patterns of employees. Feature encoding focuses on extracting more refined features from the time dimension and the space dimension. Through feature encoding, the spatial movement rules of employees in the workplace and the dynamic changes in behavior patterns can be extracted.

[0032] Figure 3 It is a block diagram schematic of the coal preparation plant employee positioning semantic encoding unit in the coal preparation plant personnel management platform based on UWB positioning according to an embodiment of the present application. As Figure 3 shown, in a specific embodiment of the present application, the coal preparation plant employee positioning semantic encoding unit 121 includes: a coal preparation plant employee positioning semantic feature extraction subunit 1211, configured to perform semantic feature extraction on the coal preparation plant employee positioning data collected by the UWB locator to obtain a coal preparation plant employee positioning vector sequence; a coal preparation plant employee positioning semantic feature understanding subunit 1212, configured to pass the coal preparation plant employee positioning vector sequence through the coal preparation plant employee positioning vector sequence semantic understanding Bi-LSTM model to obtain the coal preparation plant employee positioning text understanding feature vector.

[0033] It should be understood that the UWB positioning system can provide accurate location information, including the coordinates, movement trajectories, and stay durations of employees. However, without in-depth analysis, these raw data may not fully reflect the employees' behavior patterns or potential safety hazards at work. Through semantic feature extraction, higher-level behavioral information and semantic understanding can be extracted from this basic data to form a vector sequence with practical significance. This process first involves combining the raw positioning data with contextual information such as the employees' work environment, task requirements, and work areas to construct a data structure with deeper semantic depth. Then, natural language processing or machine learning techniques are used to model these spatial and temporal features to generate semantic feature vectors, which can express the employees' behavior patterns in different work scenarios, such as whether they are following the work process and whether there are any violations of safety regulations. In this way, a series of dynamic positioning vector sequences can be formed, with each vector containing multi-dimensional information such as time, space, and semantics of the employees' behavior. Finally, these positioning vector sequences after semantic feature extraction provide the system with richer behavioral data support, which can help managers monitor employees' activities in real time, identify potential safety hazards, and even predict possible behavioral deviations of employees, thus providing strong support for intelligent decision-making and safety management.

[0034] Furthermore, processing the coal preparation plant employee positioning vector sequence through the coal preparation plant employee positioning vector sequence semantic understanding Bi-LSTM model to obtain the coal preparation plant employee positioning text understanding feature vector is to more accurately extract high-level semantic information from the employees' behavior data, so as to better understand their working status, activity patterns, and potential safety risks. The employee positioning data collected by the UWB locator provides information in the spatial and temporal dimensions, but these raw data often lack a deep understanding of employees' behaviors and cannot directly reveal their behavioral motives or safety hazards in the plant area. By inputting these positioning vector sequences into the Bi-LSTM (Bidirectional Long Short-Term Memory Network) model, the context dependencies in the time series data can be effectively extracted, thereby enhancing the model's understanding of employees' behavior patterns. Among them, Bi-LSTM is a deep learning model that can consider both forward and backward time series information simultaneously. It can not only capture the behavioral changes of employees in the time dimension but also understand the forward and backward relevance of employees' behaviors through its bidirectional structure. For example, by inputting features such as the residence time and movement speed of employees in different areas, Bi-LSTM can simultaneously learn the behavior patterns of employees in specific time periods and specific areas, and then determine whether there are abnormal activities or potential risks. In specific operations, first, the positioning data of the coal preparation plant employees is converted into time series vectors, and each vector contains spatial position information, time information, and other context features. Then, these vector sequences are input into the Bi-LSTM model, and deep features in the time series data are extracted through a multi-layer network structure, and the dynamic changes of employees' behaviors are captured. Through the training of the Bi-LSTM model, a feature vector containing multi-dimensional information such as employees' behavior patterns and working status can be finally obtained. These feature vectors have strong semantic meanings and can effectively reflect issues such as employees' activity patterns in different situations and whether there are safety hazards.

[0035] In a specific embodiment of the present application, the coal preparation plant employee positioning semantic feature extraction sub-unit 1211 includes: passing the coal preparation plant employee positioning data collected by the UWB locator through the coal preparation plant employee positioning data semantic context encoder including an embedding layer to obtain a plurality of coal preparation plant employee positioning data semantic feature vectors; arranging the plurality of coal preparation plant employee positioning data semantic feature vectors into the coal preparation plant employee positioning vector sequence.

[0036] It should be understood that through the combination of the embedding layer and the semantic context encoder, these raw data can be transformed into more meaningful and relevant high-dimensional features, revealing the activity patterns, potential risks, and behavioral changes of employees during the work process. First, the role of the embedding layer in the semantic context encoder is to transform the original discrete location data (such as the area where the employee is located, the stay time, etc.) into a dense, low-dimensional vector representation. These embedding vectors can map the features of different locations or time periods to a shared semantic space, thereby capturing the semantic information of employees' activities in different regions and time periods. In this way, the dimension of the location data is compressed and transformed from the original spatial and temporal information into vectors with deeper semantic meanings. Next, the semantic context encoder further explores the context relationship of the data by combining the feature vectors output by the embedding layer. Through the context encoder, more complex modeling of the employees' location behaviors can be carried out to capture the behavioral patterns of employees under different time and space conditions. In this way, by using the semantic context encoder for the location data of coal preparation plant employees containing the embedding layer, the original location data can be effectively transformed into semantic-rich and highly relevant feature vectors, helping to achieve more in-depth employee behavior analysis and intelligent management. In a specific embodiment of the present application, the location data of coal preparation plant employees collected by the UWB locator is passed through the semantic context encoder for the location data of coal preparation plant employees containing the embedding layer to obtain multiple semantic feature vectors for the location data of coal preparation plant employees, including: performing word segmentation processing on the location data of coal preparation plant employees collected by the UWB locator to obtain an employee location word sequence; using the embedding layer of the semantic context encoder for the location data of coal preparation plant employees containing the embedding layer to map each employee location word in the employee location word sequence into a word embedding vector respectively to obtain a sequence of employee location word embedding vectors; using the Transformer-based Bert model of the semantic context encoder for the location data of coal preparation plant employees containing the embedding layer to perform global context semantic encoding on the sequence of employee location word embedding vectors to obtain multiple semantic feature vectors for the location data of coal preparation plant employees.

[0037] Furthermore, arranging the semantic feature vectors of the positioning data of multiple coal preparation plant employees into a positioning vector sequence of coal preparation plant employees is to integrate the scattered positioning data at a single moment into a continuous time series, so as to better analyze and understand the behavior dynamics, work patterns and potential risks of employees. Among them, the original positioning data of coal preparation plant employees usually includes information such as spatial positions and residence durations at each moment. These information have a time sequence and context relationship. After arranging them into a vector sequence, the behavior changes and activity trajectories of employees within different time periods can be captured, providing richer analysis dimensions. Among them, each vector represents the behavior state and position characteristics of an employee at a certain moment, and the entire sequence shows the activity trajectory and behavior changes of the employee within a certain time period. This arrangement method can help the system capture the time dependence of employees' behaviors. By analyzing these time series data, the activity rules of employees can be identified and potential problems can be discovered.

[0038] Specifically, arranging the semantic feature vectors of the positioning data of the multiple coal preparation plant employees into the positioning vector sequence of coal preparation plant employees includes: creating a Spring Boot project, configuring the pom.xml file, adding dependencies of relevant libraries and tools for sequence processing; creating an EmployeeLocationController class for receiving client requests and calling corresponding service layer methods; defining methods in the service layer to receive the employee positioning data obtained from the UWB locator and arrange the feature vectors; configuring a RESTful API interface for receiving data and returning the synthesized feature vector sequence; running the Spring Boot application and performing tests and validations.

[0039] Among them, part of the deployment code for creating the Spring Boot main application class is as follows.

[0040]

[0041] Among them, part of the deployment code for creating the Controller class is as follows.

[0042]

[0043]

[0044] Among them, part of the deployment code for creating the Service class is as follows.

[0045]

[0046]

[0047] Among them, part of the deployment code for creating the data model class is as follows.

[0048]

[0049]

[0050] Among them, the partial deployment code for configuring application.properties is as follows.

[0051] server.port=8080

[0052] Among them, the partial deployment code for starting the application is as follows.

[0053] Run the LocationApplication class in the IDE or use the Maven command:

[0054] bash

[0055] mvn spring-boot:run

[0056] It should be understood that the development using the Spring Boot framework aims to achieve the function of arranging the semantic feature vectors of the positioning data of employees in multiple coal preparation plants into a sequence of employee positioning vectors. Among them, the Controller layer is responsible for receiving requests from the client, processing the positioning data of employees in the coal preparation plant, and interacting with the front-end system through the RESTful API. The specific business logic is encapsulated in the Service layer. Its main function is to obtain the employee positioning data from the UWB locator, extract its semantic feature vectors, and arrange these feature vectors into a complete vector sequence according to certain rules. The data model layer defines the structure of the employee positioning data, including the employee ID and the corresponding feature vectors. Among them, the EmployeeLocationController controller is defined to handle the POST request, pass in a list of employee positioning data, and pass it to the business processing logic in the LocationService class. This logic first extracts the feature vectors of each employee positioning data, and then concatenates these feature vectors into a long vector sequence to prepare for the subsequent analysis of the deep learning model. In actual operation, the arrangeFeatureVectorsIntoSequence method arranges multiple feature vectors to generate the final vector sequence. In this way, the project can convert the raw data collected from the UWB locator into a format suitable for further analysis. In this way, it can efficiently process and arrange the positioning data of different employees, ensuring the accuracy and timeliness of subsequent anomaly detection and positioning early warning. Moreover, through the RESTful interface of Spring Boot, users or other systems can conveniently submit positioning data, and the system will automatically process and feedback the results, ensuring efficient data processing fluency and scalability, and meeting the needs of enterprises such as coal preparation plants in employee positioning, monitoring, and safety management.

[0057] In a specific embodiment of the present application, the coal preparation plant employee positioning feature encoding unit 122 includes: performing word segmentation on the coal preparation plant employee positioning data collected by the UWB locator to obtain a plurality of coal preparation plant employee positioning data items; passing the plurality of coal preparation plant employee positioning data items through the coal preparation plant employee positioning data multi-scale feature extractor to obtain the coal preparation plant employee positioning data associated feature vectors.

[0058] It should be understood that the raw data provided by the UWB positioning system contains the spatial location information, activity trajectories, timestamps, residence durations, etc. of employees. Although these data are accurate, they may be too complex to directly use and difficult to extract valuable and actionable features from. Therefore, by performing "word segmentation" on these data, they can be broken down into multiple relatively simple and independently analyzable elements, providing a clearer and more manageable data structure for subsequent behavior analysis, pattern recognition, and security monitoring. Specifically, word segmentation processing usually splits the original positioning data into independent data items according to certain rules. For example, the UWB positioning data may contain information such as time, position coordinates, moving speed, activity areas, etc., and each dimension can be regarded as a separate data item. Through the word segmentation operation, these raw data will be cut into word items or data items with individual meanings, and each data item corresponds to the behavior characteristics of employees at a specific moment or within a specific spatial range. This breakdown not only simplifies the data structure but also enables different data items to play their respective roles in subsequent analyses. For example, analyzing employees' activity areas, working hours, and area switches, etc. The multiple positioning data items obtained through word segmentation can facilitate data cleaning, feature extraction, and behavior pattern modeling more conveniently.

[0059] Furthermore, the original employee positioning data contains multiple data items, such as time, spatial location, moving speed, activity area, etc. Relying solely on data analysis at a certain scale may not be able to comprehensively capture the behavior patterns of employees in a complex environment. The multi-scale feature extractor can extract the features of data from different scales and perspectives, enabling the model to more accurately reveal the multi-dimensional correlations of employee behavior. In a specific embodiment of the present application, first, multiple coal preparation plant employee positioning data items are input into the multi-scale feature extractor. By processing data at different scales, the multi-scale feature extractor can analyze the behavior of employees at different levels, from macroscopic to microscopic and from global to local. Then, through operations such as convolution and pooling on the data at different scales, the multi-scale feature extractor can automatically extract the most meaningful correlation features in the data. These extracted features often contain important information such as the behavior patterns of employees at different time periods, the distribution of activity areas, and potential safety risks. Finally, after multi-scale processing, the obtained correlation feature vector not only reflects the overall behavior of employees but also reveals the correlations between local behaviors, thereby improving the accuracy of employee behavior prediction. In this way, multiple positioning data items are transformed into a high-dimensional correlation feature vector containing multi-level information, which can more comprehensively describe the activity status of employees in the coal preparation plant. These feature vectors are not only convenient for further analysis, modeling, and risk prediction but also can provide support for intelligent management, employee safety monitoring, work efficiency improvement, etc. Specifically, obtaining the coal preparation plant employee positioning data correlation feature vector by passing the multiple coal preparation plant employee positioning data items through the coal preparation plant employee positioning data multi-scale feature extractor includes: using each layer of the coal preparation plant employee positioning data multi-scale feature extractor to respectively perform the following operations on the input data during the forward pass of the layer: performing convolution processing on the input data based on the first convolution kernel to obtain a first convolution feature map; performing convolution processing on the input data based on the second convolution kernel to obtain a second convolution feature map; performing convolution processing on the input data based on the third convolution kernel to obtain a third convolution feature map; performing convolution processing on the input data based on the fourth convolution kernel to obtain a fourth convolution feature map, where the first convolution kernel, the second convolution kernel, the third convolution kernel, and the fourth convolution kernel have different sizes; concatenating the first convolution feature map, the second convolution feature map, the third convolution feature map, and the fourth convolution feature map to obtain a multi-scale convolution feature map; performing mean pooling processing on the multi-scale convolution feature map along the channel dimension to obtain a pooled feature map; and performing non-linear activation processing on the pooled feature map to obtain an activation feature map; where the output of the last layer of the coal preparation plant employee positioning data multi-scale feature extractor is the coal preparation plant employee positioning data correlation feature vector.

[0060] In the above-mentioned personnel management platform 100 for coal preparation plants based on UWB positioning, the personnel positioning anomaly judgment module 130 is used to judge whether to issue a personnel positioning anomaly warning alarm based on the coal preparation plant employee positioning text understanding feature vector and the coal preparation plant employee positioning data association feature vector. It should be understood that the coal preparation plant employee positioning text understanding feature vector is obtained by semantic analysis of the employee's behavior data, and these feature vectors can reflect the employee's activity patterns, behavior rules, and task completion status during work. The coal preparation plant employee positioning data association feature vector captures the dynamic changes of the employee's behavior at different time and space scales through multi-scale feature extraction, and can reveal the global and local features of the employee's behavior. When these two feature vectors are combined, the system can compare the differences between the employee's current behavior and the historical behavior pattern to judge whether there are abnormal situations. Finally, when the system identifies that the employee's behavior significantly deviates from the normal pattern, it will automatically issue a personnel positioning anomaly warning alarm to prompt the management to pay attention to the employee's safety, work status, or potential risks, and take timely intervention measures to ensure production safety and employee health.

[0061] Figure 4 FIG. is a block diagram schematic of the personnel positioning anomaly judgment module in the personnel management platform for coal preparation plants based on UWB positioning according to an embodiment of the present application. As Figure 4 shown, in a specific embodiment of the present application, the personnel positioning anomaly judgment module 130 includes: a coal preparation plant personnel positioning feature fusion unit 131, configured to fuse the coal preparation plant employee positioning text understanding feature vector and the coal preparation plant employee positioning data association feature vector to obtain a personnel positioning anomaly warning feature vector; a coal preparation plant personnel positioning feature optimization unit 132, configured to perform multi-objective matching optimization based on the target domain edge anchor on the personnel positioning anomaly warning feature vector to obtain an optimized personnel positioning anomaly warning feature vector; a personnel positioning anomaly warning alarm classification judgment unit 133, configured to pass the optimized personnel positioning anomaly warning feature vector through a classifier to obtain a classification result, and the classification result is used to judge whether to issue a personnel positioning anomaly warning alarm.

[0062] It should be understood that integrating the text understanding feature vector of coal preparation plant employees and the data association feature vector of coal preparation plant employees can comprehensively improve the accuracy and sensitivity of abnormal behavior recognition. Among them, the two feature vectors provide key information about employees' behaviors from different perspectives. The text understanding feature vector mainly reflects the activity semantics and work content of employees, while the data association feature vector captures the spatial and temporal patterns of employees' behaviors. By fusing these two feature vectors, the behavior data, location changes, and work status of employees can be effectively integrated, thus forming a more comprehensive and accurate abnormal warning feature. In the technical solution of this application, in order to fuse these two feature vectors, multiple methods can be adopted. For example, first, these two feature vectors can be concatenated in the same dimension, combining the respective information of the text understanding feature vector and the data association feature vector to form a new feature vector containing multi-dimensional information. Then, weighted average or fusion models (such as neural networks, decision trees, etc.) can be used to fuse these two feature vectors and adjust their weights in the final feature vector to adapt to different scenarios and requirements in practical applications. The fused feature vector not only contains the behavior status of employees at work (such as task completion status, regional residence time, etc.), but also integrates the spatial behavior patterns of employees (such as regional switching frequency, movement speed, etc.) and potential abnormal patterns (such as staying in non-work areas for a long time, frequently entering and leaving dangerous areas, etc.).

[0063] In particular, considering that the semantic feature vector of the coal preparation plant employee positioning data is the positioning data collected by the UWB locator, the semantic features obtained by the context encoder mainly describe the employee's behavior pattern, location sequence and other information, and have temporal and semantic richness. The employee positioning data association feature vector is the association feature extracted by the multi-scale feature extractor after the positioning data is segmented, which is usually closely related to the physical environment where the employee is located, emphasizing the association between spatial characteristics and physical environment. The two represent information in different fields, one is biased towards temporal and semantic content, and the other is more focused on spatial and environmental factors. In this context, due to the different representation spaces, dimensions and internal structures of the two types of features, there will be significant differences in feature distribution. Among them, the semantic feature vector may be closer to the distribution of text or sequence data, and may have higher nonlinearity and complex patterns, while the association feature vector is closer to the traditional spatial vector, and may be more regular in distribution and is greatly affected by environmental factors. Traditional feature fusion methods usually ignore the distribution properties of these feature domains, resulting in information conflicts and redundancies between different features not being effectively suppressed, causing overfitting or underfitting of the model, so that the fused feature space may not accurately express the complexity of the actual scene, resulting in a decrease in the generalization ability of the model. In order to improve the performance and stability of the model, the distribution differences of each feature must be considered. Therefore, in the technical solution of the present application, the personnel positioning abnormality warning feature vector is optimized by performing multi-target matching optimization based on the edge anchoring of the target domain to obtain an optimized personnel positioning abnormality warning feature vector.

[0064] Among them, the multi-objective matching optimization based on the target domain edge anchoring is performed on the personnel positioning abnormality warning feature vector to obtain the optimized personnel positioning abnormality warning feature vector, including: mapping the personnel positioning abnormality warning feature vector to the intrinsic decomposition space to obtain the first intrinsic decomposition feature vector; extracting the maximum eigenvalue and the minimum eigenvalue of the first intrinsic decomposition feature vector; calculating the difference between the maximum eigenvalue and the minimum eigenvalue as the target domain edge anchoring description operator; extracting the mean and standard deviation of the first intrinsic decomposition feature vector, and dividing the mean by the standard deviation to obtain the optimized direction description operator; based on the target domain edge anchoring description operator and the optimized direction description operator, matching optimization is performed on the personnel positioning abnormality warning feature vector to obtain the optimized personnel positioning abnormality warning feature vector.

[0065] The abnormal warning feature vector of the personnel positioning is mapped to the eigendecomposition space to obtain the first eigendecomposition feature vector, which is represented by the following mapping formula:

[0066]

[0067] Among them, V 1Denote the abnormal warning feature vector of personnel positioning, PCA(V 1 ) represents mapping V 1 into the eigen - decomposition space, U 1 is the sequence of the first eigen - decomposition vectors, Λ 1 is the first diagonal matrix, U 1 T is the transpose of U 1 , v 11 , v 12 , v 1m are the first, second and the m - th eigen - vectors of the sequence of the first eigen - decomposition vectors, λ i1 , λ im are the eigenvalues at the first position and the m - th position of the first diagonal matrix respectively, and V represents the first eigen - decomposition feature vector.

[0068] Among them, the abnormal warning feature vector of personnel positioning is optimized by multi - target matching based on the edge anchoring of the target domain to obtain the optimized abnormal warning feature vector of personnel positioning, which is represented by the following optimization formula:

[0069]

[0070] Among them, v max and v min represent the maximum eigenvalue and the minimum eigenvalue of the first eigen - decomposition feature vector respectively, α represents the target - domain edge - anchoring description operator, μ and σ represent the mean and standard deviation of the first eigen - decomposition feature vector respectively, τ represents the optimization - direction description operator, ⊙ and represent point - by - point addition, point - by - point multiplication and point - by - point subtraction respectively, exp represents the natural exponential function with the natural constant e as the base, V 1 ⊙-1 represents calculating the reciprocal of each eigenvalue of the abnormal warning feature vector of personnel positioning, V 1 represents the first eigen - decomposition feature vector, and V' represents the optimized abnormal warning feature vector of personnel positioning.

[0071] In the technical solution of this application, multi-objective matching optimization based on target domain edge anchoring is performed on the feature vector of abnormal warning for personnel positioning. This process first maps the feature vector of abnormal warning for personnel positioning to the eigen-decomposition space. By means of eigen-decomposition (such as classical numerical algebraic methods like eigenvalue decomposition, singular value decomposition, etc.), the original high-dimensional feature vector is projected onto a low-dimensional space composed of orthogonal bases. The directions in this low-dimensional space are defined by eigenvectors, and the corresponding eigenvalues characterize the variance or intensity distribution of the data in different directions. The process of projecting onto the eigen-decomposition space not only removes the redundant noise in the original features through dimensionality reduction but also maps the original vector with a complex distribution to a physically more intuitive decomposition dimension, making the structural information of the features clearer.

[0072] After the mapping is completed, it is necessary to further extract the maximum eigenvalue and the minimum eigenvalue of the first eigen-decomposition feature vector. These two values respectively reflect the data characteristics of the feature vector in the most significant direction and the least significant direction. Among them, the maximum eigenvalue represents the proportion of data information in the main direction and is the core explanatory factor for the feature distribution, while the minimum eigenvalue is usually related to noise or data errors and represents the weakest change in the direction. The maximum and minimum eigenvalues are important representations of the data space form. In the analysis and optimization process, they provide a mathematical description of the global distribution characteristics of the features. At the same time, the distribution pattern of the eigenvalues also implies the complexity of the data in the target domain. For example, when the maximum eigenvalue is much larger than other eigenvalues, the feature has an obvious main axis direction in this space; if the eigenvalue distribution is relatively uniform, there may be higher complexity or diversity.

[0073] In order to construct further feature optimization indicators, it is necessary to calculate the difference between the maximum eigenvalue and the minimum eigenvalue and define it as the target domain edge anchoring description operator. The eigenvalue difference clearly defines the distribution range and span of the feature in the target domain from a geometric perspective. It corresponds to the difference between the major axis and the minor axis in the eigen-decomposition dimension and can be regarded as an anchoring index for the feature distribution, reflecting the boundary characteristics that the feature may exhibit in the target domain. The theoretical significance of the edge anchoring description operator is not limited to describing the data distribution difference. It also provides a measurement method for capturing the information specific points that establish the boundary performance in the feature space and supports the subsequent process of dividing and adjusting the target domain features. Through deep association with the target domain, this operator can further guide the subsequent optimization steps, making the optimized feature vector closely adhere to the domain edge attributes and enhancing its in-domain adaptability.

[0074] Meanwhile, it is also necessary to extract the mean and standard deviation from the first eigen-decomposition eigenvector, and calculate their ratio to obtain the optimized direction description operator. The mean characterizes the central tendency of the eigenvalues in a specific direction, while the standard deviation describes their discrete characteristics within the overall range. By normalizing the ratio of the mean to the standard deviation, this optimized direction description operator provides a normalization mechanism, enabling the feature optimization process to better handle the imbalance of different data scales while ensuring the robustness of the model. In the high-dimensional feature space, the ratio of the mean to the standard deviation further provides a stable basis for direction optimization, ensuring that the optimization process is not overly disturbed by extreme values or outliers. At the same time, this operator can also be understood as an adaptive adjustment rule for direction selection, making the optimization result more conform to the inherent characteristics of the target domain, especially achieving a balance between direction alignment and noise suppression.

[0075] After the construction of the above key operator, the operation of finally matching and optimizing the abnormal warning feature vector of personnel positioning comprehensively uses the target domain edge anchoring description operator and the optimized direction description operator to improve the discrimination ability of the feature vector and obtain an optimized representation. Matching optimization is a multi-objective optimization strategy that requires balancing the edge attributes and directional attributes of the feature distribution simultaneously to avoid overfitting or deviation caused by preferring a certain objective. The edge anchoring description operator ensures that the distribution of the feature vector can better reflect the global and local distribution patterns of the target domain by capturing the significance of the edge characteristics; the optimized direction description operator further provides a reference for direction adjustment in this process, thus ensuring that the optimization can proceed in a more effective gradient direction.

[0076] Furthermore, the optimized feature vector for personnel location anomaly warning includes multi-dimensional behavioral data of employees in the workplace, including spatial location, residence time, area switching frequency, etc. The feature vector reflects the overall behavioral state of employees and potential abnormal patterns, which may involve abnormal behaviors such as employees staying in dangerous areas, stagnating at a certain position for a long time, and frequently changing work areas. The classifier constructs a model to identify whether the current behavioral pattern conforms to the normal work process by learning behavioral samples labeled as "normal" or "abnormal" in historical data. Among them, the working principle of the classifier is usually trained through supervised learning methods. In the training stage, the system inputs the personnel location data labeled as normal and abnormal according to historical data, and establishes a model through classification algorithms (such as support vector machines, decision trees, random forests, neural networks, etc.). The model makes classification judgments based on the features of each feature vector, learns and identifies which features represent "normal" behaviors and which features may mean "abnormal" behaviors. In this way, the classifier can make judgments for new input data. When the new optimized feature vector for personnel location anomaly warning is input into the classifier, the classifier will classify it as "normal" or "abnormal" according to the content of the feature vector. If the classification result is "abnormal", the system will trigger a personnel location anomaly warning alarm, and notify the management personnel or safety monitoring personnel in time for further processing, such as checking the safety status of employees and determining whether there are potential risks. On the contrary, if the classification result is "normal", it means that the employee's behavior is within the normal range and no further intervention is required. In this way, the classifier can achieve fast and accurate detection of abnormal behaviors, greatly improving the efficiency of personnel safety monitoring. The system can respond to employees' abnormal behaviors in real time, give early warnings, prevent potential safety accidents or work oversights, provide strong data support for the management personnel of the coal preparation plant, and ensure the safe and efficient operation of the production environment.

[0077] In summary, the embodiment of this application first obtains the personnel location data of the coal preparation plant employees collected by the UWB locator, then uses deep learning technology to extract features and perform correlation analysis on it, and finally uses a classifier to determine whether to issue a personnel location anomaly warning alarm, so as to effectively prevent potential safety accidents, reduce management costs, and further improve the safety, production efficiency and management level of the coal preparation plant employees.

[0078] As described above, the personnel management platform 100 for coal preparation plants based on UWB positioning according to the embodiments of the present application can be implemented in various terminal devices. In one example, the personnel management platform 100 for coal preparation plants based on UWB positioning can be integrated into the terminal device as a software module and / or a hardware module. For example, the personnel management platform 100 for coal preparation plants based on UWB positioning can be a software module in the operating system of the terminal device, or can be an application program developed for the terminal device; of course, the personnel management platform 100 for coal preparation plants based on UWB positioning can also be one of the numerous hardware modules of the terminal device.

[0079] Alternatively, in another example, the personnel management platform 100 for coal preparation plants based on UWB positioning and the terminal device can also be separate devices, and the personnel management platform 100 for coal preparation plants based on UWB positioning can be connected to the terminal device through a wired and / or wireless network, and transmit and interact information according to a predefined data format.

[0080] Figure 5 FIG. is a flowchart of personnel management for coal preparation plants based on UWB positioning according to the embodiments of the present application. As Figure 5 shown, the personnel management for coal preparation plants based on UWB positioning according to the embodiments of the present application includes: S110, obtaining the positioning data of coal preparation plant employees collected by the UWB locator; S120, extracting the text understanding feature vector of the positioning of coal preparation plant employees and the correlation feature vector of the positioning data of coal preparation plant employees from the positioning data of coal preparation plant employees collected by the UWB locator; S130, based on the text understanding feature vector of the positioning of coal preparation plant employees and the correlation feature vector of the positioning data of coal preparation plant employees, determining whether to issue an alarm for abnormal personnel positioning.

[0081] Here, those skilled in the art can understand that the specific operations of each step in the above-mentioned personnel management for coal preparation plants based on UWB positioning have been described in detail in the description of the personnel management platform for coal preparation plants based on UWB positioning above, and therefore, the repeated description thereof will be omitted. Figures 1 to 4 In several embodiments provided by the present invention, it should be understood that the disclosed devices, apparatuses and methods can be implemented in other ways. For example, the device embodiments described above are merely illustrative. For example, the division of the modules is only a logical function division, and there may be other division methods in actual implementation.

[0082]

[0083] ​The module described as a separation component may or may not be physically separated. The component shown as a module may or may not be a physical unit, that is, it may be located in one place or distributed to multiple network units. Some or all of the modules can be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0084] In addition, in each embodiment of the present invention, each functional module can be integrated in a processing unit, or each unit can exist physically alone, or two or more units can be integrated in one unit. The above integrated unit can be implemented in the form of hardware or in the form of a combination of hardware and software functional modules.

[0085] For those skilled in the art, it is obvious that the present invention is not limited to the details of the above exemplary embodiments, and without departing from the spirit or basic characteristics of the present invention, the present invention can be implemented in other specific forms.

[0086] Therefore, from any point of view, the embodiments should be regarded as exemplary and non-restrictive. The scope of the present invention is defined by the appended claims rather than the above description. Therefore, all changes falling within the meaning and scope of the equivalent elements of the claims are intended to be included in the present invention. Any reference signs in the claims should not be regarded as limiting the claimed rights.

[0087] In addition, it is obvious that the word "comprising" does not exclude other units or steps, and the singular does not exclude the plural. The multiple units or devices stated in the system claims can also be implemented by one unit or device through software or hardware. Words such as "second" are used to denote names and do not denote any particular order.

[0088] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit them. Although the present invention has been described in detail with reference to the preferred embodiments, those of ordinary skill in the art should understand that the technical solutions of the present invention can be modified or equivalently replaced without departing from the spirit of the technical solutions of the present invention.

Claims

1. A coal preparation plant personnel management platform based on UWB positioning, characterized in that: include: The coal preparation plant employee data acquisition module is used to obtain the coal preparation plant employee positioning data collected by the UWB locator; A coal preparation plant employee data extraction module, used to extract a coal preparation plant employee positioning text understanding feature vector and a coal preparation plant employee positioning data association feature vector from the coal preparation plant employee positioning data collected by the UWB locator; The personnel positioning abnormality judgment module is used to judge whether to issue a personnel positioning abnormality warning alarm based on the coal preparation plant employee positioning text understanding feature vector and the coal preparation plant employee positioning data association feature vector.

2. The coal preparation plant personnel management platform based on UWB positioning according to claim 1 is characterized in that: The coal preparation plant employee data extraction module includes: A coal preparation plant employee positioning semantic encoding unit, used for performing semantic encoding on the coal preparation plant employee positioning data collected by the UWB locator to obtain the coal preparation plant employee positioning text understanding feature vector; The coal preparation plant employee positioning feature encoding unit is used to perform feature encoding on the coal preparation plant employee positioning data collected by the UWB locator to obtain a feature vector associated with the coal preparation plant employee positioning data.

3. The coal preparation plant personnel management platform based on UWB positioning according to claim 2 is characterized in that: The coal preparation plant employee positioning semantic coding unit includes: A coal preparation plant employee positioning semantic feature extraction subunit is used to extract semantic features from the coal preparation plant employee positioning data collected by the UWB locator to obtain a coal preparation plant employee positioning vector sequence; The coal preparation plant employee positioning semantic feature understanding subunit is used to pass the coal preparation plant employee positioning vector sequence through the coal preparation plant employee positioning vector sequence semantic understanding Bi-LSTM model to obtain the coal preparation plant employee positioning text understanding feature vector.

4. The coal preparation plant personnel management platform based on UWB positioning according to claim 3 is characterized in that: The coal preparation plant employee positioning semantic feature extraction subunit includes: Passing the coal preparation plant employee positioning data collected by the UWB locator through a coal preparation plant employee positioning data semantic context encoder including an embedding layer to obtain a plurality of coal preparation plant employee positioning data semantic feature vectors; The plurality of coal preparation plant employee location data semantic feature vectors are arranged into the coal preparation plant employee location vector sequence.

5. The coal preparation plant personnel management platform based on UWB positioning according to claim 4 is characterized in that: Arranging the plurality of coal preparation plant employee location data semantic feature vectors into the coal preparation plant employee location vector sequence comprises: Create a Spring Boot project, configure the pom.xml file, add dependencies of related libraries and tools for sequence processing; Create the EmployeeLocationController class to receive client requests and call corresponding service layer methods; Define methods in the service layer to receive employee location data obtained from the UWB locator and arrange the feature vectors; Configure a RESTful API interface to receive data and return the synthesized feature vector sequence; Run the Spring Boot application and perform testing and verification.

6. The coal preparation plant personnel management platform based on UWB positioning according to claim 5 is characterized in that: The coal preparation plant employee positioning feature coding unit comprises: Segmenting the coal preparation plant employee positioning data collected by the UWB locator to obtain a plurality of coal preparation plant employee positioning data items; The multiple coal preparation plant employee positioning data items are passed through a coal preparation plant employee positioning data multi-scale feature extractor to obtain the coal preparation plant employee positioning data associated feature vectors.

7. The coal preparation plant personnel management platform based on UWB positioning according to claim 6 is characterized in that: The personnel positioning abnormality judgment module includes: A coal preparation plant personnel positioning feature fusion unit, used to fuse the coal preparation plant employee positioning text understanding feature vector and the coal preparation plant employee positioning data association feature vector to obtain a personnel positioning abnormality warning feature vector; A coal preparation plant personnel positioning feature optimization unit is used to perform multi-objective matching optimization based on target domain edge anchoring on the personnel positioning abnormality warning feature vector to obtain an optimized personnel positioning abnormality warning feature vector; The personnel positioning abnormality warning alarm classification judgment unit is used to pass the optimized personnel positioning abnormality warning feature vector through a classifier to obtain a classification result, and the classification result is used to determine whether to issue a personnel positioning abnormality warning alarm.

8. The coal preparation plant personnel management platform based on UWB positioning according to claim 7 is characterized in that: The coal preparation plant personnel positioning feature optimization unit comprises: Mapping the personnel positioning abnormality warning feature vector to an eigendecomposition space to obtain a first eigendecomposition feature vector; Extracting the maximum eigenvalue and the minimum eigenvalue of the first eigendecomposition eigenvector; Calculate the difference between the maximum eigenvalue and the minimum eigenvalue as a target domain edge anchoring description operator; Extracting a mean value and a standard deviation of the first eigendecomposition eigenvector, and dividing the mean value by the standard deviation to obtain an optimized direction description operator; Based on the target domain edge anchoring description operator and the optimization direction description operator, the personnel positioning abnormality warning feature vector is matched and optimized to obtain the optimized personnel positioning abnormality warning feature vector.

9. A coal preparation plant personnel management based on UWB positioning, characterized in that: include: Obtain the coal preparation plant employee location data collected by the UWB locator; Extracting a coal preparation plant employee positioning text understanding feature vector and a coal preparation plant employee positioning data association feature vector from the coal preparation plant employee positioning data collected by the UWB locator; Based on the coal preparation plant employee positioning text understanding feature vector and the coal preparation plant employee positioning data association feature vector, it is determined whether to issue a personnel positioning abnormality warning alarm.

10. The personnel management of a coal preparation plant based on UWB positioning according to claim 9, characterized in that: Extracting the coal preparation plant employee positioning text understanding feature vector and the coal preparation plant employee positioning data association feature vector from the coal preparation plant employee positioning data collected by the UWB locator, including: Semantically encoding the coal preparation plant employee positioning data collected by the UWB locator to obtain the coal preparation plant employee positioning text understanding feature vector; The coal preparation plant employee positioning data collected by the UWB locator is feature encoded to obtain a feature vector associated with the coal preparation plant employee positioning data.

Images